This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
L > R
L < R
so that the borrowing country repays on schedule in the good state but not in the bad state. Then the expected repayment, prior to revelation of the state of the world, is 1
R P A
The World Bank, conditionality and the CDF
A claimed comparative advantage in enforcement translates into higher expected repayments as the consequence of (a) a higher threatened penalty, P , in the event of a possible default and/or (b) the imposition of greater access costs, A , subsequent to default. Now suppose the lender is able to impose policy conditionality at the time the loan is negotiated. The objective of conditionality is to advance reforms which should, over time, raise the country's income. In the willingness-to-pay framework re¯ected in models of the sort considered here, higher income levels can affect expected repayments only by altering the penalty level, P , or the cost of future access, A , in the poor state. Suppose that P is rising in income, which is likely to be the case if asset expropriation or trade inhibition are the penalties envisaged. The consequence is then that conditionality, if complied with, will raise expected repayments. In this model, conditionality therefore strengthens enforcement.3 This argument is independent of the form conditionality takes and the way that it is applied, although certain forms of conditionality and modes of application may be more effective than others. In this framework, conditionality raises expected repayments both to the World Bank, as the enforcer of conditionality, but also to private sector banks. It therefore generates an externality. The World Bank's comparative advantage in enforcement arises through internalising this externality, both because it values development per se and because it values governments servicing private sector borrowing in order to maintain future access to credit markets. Private sector banks will be concerned only about their own repayments, and indeed may even try to obtain these at the expense of repayments to other banks (through assertion of seniority). Paradoxically, therefore, by asserting a development objective in addition to a pro®t objective, the Bank's lending activities may result in the selection of the cooperative equilibrium. If this does happen, both private banks and the Bank itself may enjoy a higher level of income than that which would have resulted if the Bank had been concerned solely with pro®ts. This raises the question of whether a market comprising purely private banks could also attain the cooperative equilibrium. This is a possible reformulation of Rodrik's (1995) question, `Why is there multilateral lending?' However, the World Bank, as the IBRD, lends against its capital by borrowing from ®nancial markets rather than lending the capital directly so the de®ning feature of multilateral lending is the enforcement of repayment rather than the source of funds. Multilateral lending is `multilateral' in the sense that borrowing and lending governments both accord it preferred status. Could preferred status ever describe an equilibrium in a market of purely private sector institutions?
R. Hopkins, A. Powell, A. Roy and C. L. Gilbert
Any individual pro®t-maximising bank will have an incentive to defect in bad states by placing repayment on its own loans ahead of the general development objective which would generate a higher overall level of repayment. The cooperative equilibrium can be sustained only if banks can credibly commit not to seek repayment on a unilateral basis. Only multilateral institutions can make such commitments, which are effectively guaranteed by their charters and by their ownership structures. It matters that the Bank's shareholders are governments and not investors, and that, furthermore, a proportion of the voting equity is held by borrowing governments. These conceptual discussions are important because different views of the fundamental economic functions which the World Bank performs will have different implications for current policy debates. We draw two lessons from the foregoing discussion: (1) Because of its ability to apply policy conditionality, the World Bank is complementary with private sector development lending. Better policy will enhance lending opportunities and potential rates of return and result in superior debt service. (2) This externality would not be available in the absence of a strong public sector institution which combines banking and development functions.4
4 The conditionality debate There has been signi®cant recent research on the role and effectiveness of conditionality (summarised in World Bank 1998; see also Killick, 1998 and Deverajan and Swaroop and Burnside and Dollar, chapters 7 and 8 in this volume). Much of this work has been conducted considering conditionality with respect to aid ¯ows or highly subsidised lending. However, while we recognise that conditionality may take various forms (and in particular there has been a marked difference in IMFstyle versus World Bank-style conditionality), we consider this analysis as highly relevant not only for highly subsidised lending, but also for IBRD-type ¯ows. The prevalent view from this literature is that aid is effective but only in conjunction with `good' policy. As a result there is a movement towards favouring `policy-level conditionality' and away from `policy-change conditionality'.5 The implication is that, to obtain good value for each dollar spent, donors should be much more selective as to which countries they provide aid to.
The World Bank, conditionality and the CDF
Governments that persist in pursuing poor or weakly implemented economic policies will bene®t instead from the Bank's development advice. This approach emphasises the importance of policy ownership, and of dialogue and partnership between donors and recipients. Another useful distinction has been proposed by Killick (1998), between hard core conditionality and pro forma conditionality. The former refers to `a requirement for involuntary action of some sort on the part of the recipient government without which assistance would not be granted or continued' (Killick, 1998: 27). We prefer the terms `imposed' and `agreed' conditionality, on the basis that Killick's nomenclature may be thought pejorative, and, in particular, because the term `pro forma' might give the incorrect impression that these conditions are not serious. Agreed conditionality relates to policy commitments included in aid agreements for the convenience of both parties (Killick, 1998: 188). They are measures laying down what could be done and in what sequence, serving as a kind of institutional memory against the possibility of changes among key of®cials. In general, Killick refers to `conditionality' as the imposed , hard core component of it, although it admits that there is a grey area between the two types. Although the debate is far from over, a number of conclusions have been reached. Perhaps the most important one is that conditionality, at least in the form that has been implemented, has had very limited results. The explanations for this are various: (1) the overriding importance of domestic political factors as determinants of long-run policy changes; (2) the dif®culties of international organisations to put in place a consistent and credible set of rewards and penalties; (3) serious problems of implementation; and (4) the in¯uence of external shocks, making it dif®cult for governments to keep to their original commitments. Perhaps the most interesting aspect in the debate on conditionality relates to the search for an alternative. For Collier (chapter 12 in this volume) there are a number of factors that create an opportunity to redesign conditionality (the end of the Cold War, the concerted efforts to reduce indebtedness, the improvement in economic policies in many developing countries and the results of recent research on aid effectiveness). Collier argues that, aside from exceptional circumstances, the relationship between aid recipients and donors should be viewed as a partnership between a government with satisfactory policies and aid providers. In this approach, conditionality is not abandoned but redesigned: `I am thus suggesting that donor conditionality should, to be effective, take a different form . . . donors should condition their ®nancial resource transfers not upon policy change but upon policy levels.' To other authors, the new role for conditionality is not so clear.
R. Hopkins, A. Powell, A. Roy and C. L. Gilbert
Killick (1998) suggests continuing using agreed conditionality but rejects the imposed variety. He is willing to accept the use of imposed conditionality only under exceptional circumstances (e.g. where there are speci®c grounds for believing that levying policy conditions will tip the political balance in favour of change, 1998: 190). He proposes, instead, an alternative model of donor±recipient relations, based upon four principles: ownership, selectivity, support and dialogue. The contribution of Killick is in this respect particularly important as it is based on a rich body of evidence and years of direct experience advising governments of developing countries and international organisations. There is little to disagree with regarding these principles: they are a healthy reaction to an overuse of conditionality, neglecting essential aspects of the donor±recipient relationship (ownership, in particular). The question, however, is to what extent these principles imply a move away from conditionality. The principle of `selectivity' deserves a close scrutiny. It states that programme aid should be reserved for countries that have adopted for themselves ef®cient pro-development policies. The implementation of higher selectivity would increase the effectiveness of aid and the credibility of the programmes with which it is associated. In this view, donors take on responsibility only for something they can deliver. This principle implies, however, a partial reinstatement of conditionality, in that for a government to receive aid it must deliver sound policies. It has the additional complication of requiring a clear de®nition of a good-policy environment, and the identi®cation of objective indicators of ownership, a problem of which Killick is aware (1998: 180, 183±6). Selectivity therefore appears to be a form of policy level or ex ante conditionality. Non-governmental organisations (NGOs) have been quick to recognise the weakness in the argument and have raised their concern about this new form of conditionality, `rather than withdrawing from conditionality, selectivity could in reality imply more up-front conditionality' (Wood and Lockwood, 1999: 1). Recent events suggest that this is indeed not only a relevant conceptual discussion but also that it has implications for the allocation of aid.6
5 Conditionality trade-offs At least in a weak form, conditionality is inevitable, as lenders will wish to ensure repayment. World Bank lending has always involved an element of conditionality in recognition of the fact that there is little merit in
The World Bank, conditionality and the CDF
lending to ®nance a project if the potential bene®ts arising from the project are dissipated through upstream or downstream inef®ciencies. However, the major shift that resulted from the debt crisis of the 1980s was away from project-based aid to programme-based aid. In its ®rst 25 years, the Bank used conditionality in order to ensure the success of projects and to guarantee that their potential bene®ts were captured. Since the mid-1980s, lending has often been justi®ed in terms of the bene®ts of the policies adopted as the result of conditionality clauses. The policies have become the projects, with investment in economic infrastructure replacing investment in physical infrastructure. Loans are justi®ed by policy changes, instead of vice versa. If conditionality cannot be avoided, at least as one ingredient of development programmes, it is worth exploring the advantages and disadvantages of its various forms and the trade-offs involved. Table 11.1 illustrates the advantages and disadvantages of the three forms of conditionality contrasted above. Policy-change conditionality is applicable to any country but has apparently been only modestly effective, as the consequence of time-consistency problems, high monitoring costs and lack of reform ownership. It is suggested that policy-level conditionality has the potential to be much more effective, but its coverage is limited to those countries with `good' policy environments. It therefore implies a cut in the resources ¯owing to those countries whose governments fail to satisfy these conditions. It may also create a vicious circle in which negative policy environments are reinforced (particularly in those countries that are far from the minimum policy levels required for receiving aid). In the absence of the aid `bribe', governments in such countries may perceive little incentive towards implementing reforms which impose high costs in the short to medium term. The implications of altered allocation criteria for poverty alleviation are unclear. Advocates of policy-level conditionality argue that aid is fungible (Devarajan and Swaroop, chapter 7 in this volume) and, by implication, targeting aid at poverty reduction is effective only where governments embrace this objective. Against this view, it may be argued that if good policy is positively correlated with economic success (and one must expect this to be the case eventually), a decision to direct aid away from countries where governments have poor policies entails a lower priority for poverty, at least over the short to medium term. Table 11.1 emphasises that, in modifying aid allocation criteria, the Bank faces a potentially acute choice between aid effectiveness and wide aid coverage, and that the dif®culties in making this choice may be exacerbated by poverty considerations. But, in any case, the Bank may be politically constrained. Its structure ensures that all member countries are
R. Hopkins, A. Powell, A. Roy and C. L. Gilbert
Table 11.1 Forms of conditionality
a Policy-change conditionality b Policy-level conditionality
c Agreed conditionality
High (applicable to any country) Low (restricted to countries with `good-policy environments') Uncertain (depends on the donor±recipient relationship)
represented at Board level, although weighted voting may imply that a borrowing country's voice is associated with relatively little power. This at least partially representative structure may make it dif®cult for the Bank to fully implement a shift towards policy-level conditionality. Agreed conditionality constitutes a third alternative. As explained above, this type of conditionality refers to consensual policy measures agreed between donors and recipient governments. It has a great potential in terms of effectiveness and poverty alleviation (as it is correlated with ownership, Killick 1998: 11) but its scope is uncertain. It will depend on the compatibility of objectives between donors and recipient governments and on the possibility of reaching genuine consensual agreements. Both the conditionality debate, and table 11.1 that attempts to outline some of the trade-offs involved, oversimplify the practical decisions which the Bank will face in attempting to move towards a policy-level conception of conditionality. This oversimpli®cation is reminiscent of Sellars and Yates' famous pastiche of English history, 1066 and All That, in which every monarch was either a `good king' or a `bad thing'. Most developing country governments will defy this easy classi®cation. One alternative might be to relate the degree of `policy-level' conditionality to the type of `product' offered by the Bank. The current Bank's policy spectrum ranges from (1) highly subsidised lending (IDA), (2) lending to governments through the standard loan products of the IBRD, (3) lending to the private sector with government guarantees, (4) lowering the cost of ®nance through the use of guarantees either to the public or private sector, (5) lending or taking equity stakes in private
The World Bank, conditionality and the CDF
sector enterprises through the IFC and to (6) providing trade and political risk guarantees through MIGA. One possibility might be to apply different types of conditionality to different products. For example, it might be argued that policy-level conditionality might be applied more to the more advanced `products' in the World Bank product spectrum, while more traditional conditionality might still apply to the more traditional lending operations. However, the Bank Group faces greater competition in selling its more advanced (and less subsidised) products and so has less scope to put conditions on purchase of these products, and it may also be argued that it is the traditional subsidised products where increased selectivity is most important. Actual aid decisions will, more often that not, involve judgements about countries where policies are a mixture of good and less good, and implementation is a mixture of more and less effective. In these instances, it will be dif®cult to avoid augmenting level judgements with views about the direction and velocity of change, and of conditioning upon such changes. It is common to all sides in this debate that the Bank should encourage the development of good policy. In countries with `middling' policy, the Bank should aim to become a partner in working to improve policy. Aid and advice are both elements in the partnership package that the Bank can offer. However, the terms on which these are available need to be clearly understood by both sides. It will not help to pretend there is no policy-change conditionality if in fact there is an element of it in aid decisions see Eade (1997), who argues that any `fudging' of the nature of the relationship between two partners undermines the basis for honest negotiation.
6 Conditionality and the CDF The `Comprehensive Development Framework' (CDF) concept was a signi®cant development in Bank thinking during 1999 (Wolfensohn, 1999). The CDF approach to development attempts to be comprehensive in two respects: across different aspects of development, and across participants in the development process. Governments are expected, in conjunction with donors, NGOs, the private sector and `civil society', to construct a tabular CDF matrix, the cells of which indicate the required actions for each development objective (across columns) by each organisation (down rows). It is intended that construction of this matrix will
R. Hopkins, A. Powell, A. Roy and C. L. Gilbert
ensure both that development planning is comprehensive and that leadership and implementation responsibilities are clear. Bank country programmes presumably aimed for comprehensiveness across development objectives even before the CDF. However, the Bank cannot do everything, so increased coordination among donors can ensure complementarity in the development effort within an overall balanced approach. Comprehensiveness across participants is more important in that government, the private sector and `civil society' are among the intended participants.7 This implies that ownership, free markets and democracy are central to the CDF concept. The CDF has important implications for the conditionality debate which has hitherto largely been conducted among academic and Research Department economists. The CDF concept looks to substitute a cooperative and comprehensive view of development for the approach in which development objectives were imposed by donors. From the CDF standpoint, the continuing debate about the form of conditionality appears as a throwback to the previous confrontational style of donor operation. Mutually agreed targets replace unilaterally imposed conditions, and, in that sense, the CDF may be thought of as adopting Killick's concept of agreed conditionality. The comprehensive, democratic and consultative elements of the CDF process are seen as giving the targets increased legitimacy and as delivering national ownership. Increased ownership may be seen as justifying greater ¯exibility on the Bank's part ± if there is consensus about objectives, there is less requirement to impose a rigid view on means. Many of the same problems which have arisen in relation to conditionality will reappear with the CDF framework. We see two broad areas of dif®culty. First, the Bank has acknowledged that the CDF approach should result in greater selectivity in Bank funding within countries, as the consequence of increased donor coordination. Implicitly, there will also be selectivity between countries, and in this regard, priorities imposed by the CDF framework may con¯ict with selectivity based on policy quality. Will a well formulated CDF be a (or even the) criterion of satisfactory ex ante policy? Will the absence of a CDF imply that policy is unsatisfactory even if traditional indicators (openness, low in¯ation, etc.) imply sound policy? The root of this problem is that CDF targets translate into Killick's `agreed conditionality', but this may run counter to the `value for development dollars' approach which emphasises current policy quality. The second potential problem is that agreed targets can be missed just as, hitherto, conditions have failed to be met. The overall CDF philosophy is that development is a management problem. The CDF is a man-
The World Bank, conditionality and the CDF
agement tool to ensure that everyone is `on board' with agreed development objectives. There are no con¯icts of interest and no principal±agent problems. This contrasts with economists' views of development which emphasise the requirement for incentives to be correctly aligned, for clearly de®ned and enforced property rights and for measures to ameliorate malfunctioning markets. Neither the economists' nor the management view is entirely correct, and it is probably right that the balance be redressed towards better management, but exclusive reliance on better management to achieve development is likely to result in disappointment.
7 Conclusions The World Bank has evolved over its 50 years of operation so that it simultaneously exercises a number of different functions. Its core functions are those of bank and development agency. We argue that the role of conditionality is important to understanding the World Bank's operations. The Bank applies conditionality as part of its development mission ± structural adjustment has resulted in policy reforms taking priority over projects as a justi®cation for lending. But the ability to apply conditionality gives the Bank a comparative advantage in enforcement of debt service. This generates a complementarity between the Bank's development agency and banking functions. However because successful conditionality will also result in improved service of private debt, there is an externality. The Bank is concerned with development and not simply with pro®ts, and is therefore willing to allow the gains from conditionality to be shared with private sector banks. This results in a superior lending equilibrium than would exist in the absence of a strong multilateral institution. This view of the World Bank sees its comparative advantage over other ®nancial and developmental agencies as arising from the complementarity of the banking and development agency functions, which, through the application of policy conditionality, allows the internalisation of what would otherwise be a positive externality. But this view neither requires an imposed de®nition of conditionality nor that the Bank con®ne itself to this form of lending. Furthermore, there may be ample scope for improvements in implementation, or for alteration of the content of conditionality requirements to put greater emphasis on poverty alleviation or environmental concerns. In the spirit of the Comprehensive Development Framework initiative, conditionality should be seen as
R. Hopkins, A. Powell, A. Roy and C. L. Gilbert
part of the dialogue between the World Bank and recipient governments and not as an alternative to dialogue. NOTES
This is an expanded and revised version of Hopkins et al. (1997) which was itself a modi®ed version of Gilbert et al. (1996) prepared for the American Economic Association 1996 meeting in San Francisco. We are grateful to Mark Baird, Kalyan Banerji, Bejoy Das Gupta, Ron Duncan, Huw Evans, Richard Frank, Andrew Hughes Hallett, Ravi Kanbur, Tony Killick, John Mitchell, Paul Mosley, Guy Pfeffermann, Brian Pinto, Lex Rieffel, David Vines and Gerald West for useful comments on one or more of this sequence of papers. However, the views expressed are entirely our own and do not necessarily represent any of the institutions or persons listed above. Financial support comes from the Global Economic Institutions programme of the ESRC to whom we express our thanks. 1 Naim (1994) also emphasises the multiplicity of the Bank's functions but adopts a slightly different classi®cation: the `Bank-as-bank', the Bank as an instrument for `promotion of values not readily accepted by the traditional power structures within developing countries', the Bank as development consultant and the Bank as a resource-transfer mechanism. One might also see the Bank as also possessing a credit-rating function. 2 Gilbert, Powell and Vines (chapter 2 in this volume) summarises the empirical evidence. Any move towards more effective conditionality, either by making it more appropriate or even just simpler (Mosley, Harrigan and Toye, 1995), or by revising the manner in which it is applied (Collier, chapter 12 in this volume), will increase the Bank's enforcement powers. 3 Rodrik (1995) makes a similar argument that multilateral lending is justi®ed by conditionality. However, he does not go on to note that this creates a public good. Mosley (1987) and Mosley, Harrigan and Toye (1995) argue that Bank conditionality should be seen as a bargaining process between the Bank and the borrowing government, and this adds a further strand to the standard debt game. That view is developed in Fafchamps (1996), who sees conditionality as a partial substitute for the inability of sovereign borrowers to commit on repayment. However, his model, which focuses on the balance between production of traded and non-traded goods, is better seen as a model of bargaining over the extent of exchange rate overvaluation than over reform. The simple model we have used suffers from the usual limitations of one-shot models. In principle, these might be overcome in a repeated-game framework. 4 In Gilbert et al. (1996), we extended these arguments to demonstrate why then topical proposals to privatise the World Bank were misconceived (Walters, 1994; Eberstadt and Lewis, 1995). 5 Note that this discussion has also been prominent in the IMF regarding the Contingent Credit Line (CCL), albeit with different language. The now approved CCL explicitly includes `policy-level' conditionality or ex ante preconditions in the language of the Fund. The idea is that this more automati-
The World Bank, conditionality and the CDF
cally disbursing credit line, which is designed to limit contagion from one emerging country with economic problems to another with sound fundamentals, will be available only to countries that satisfy certain policy conditions. 6 A note in the March 1999 Bretton Woods Update, entitled `Selectivity: conditionality by any other name', summarises the meeting between government of®cials and the World Bank discussing new criteria for allocating its IDA resources among countries. Participants made a number of critical comments on the 20 criteria describing a `good-policy environment', asserting, among other things that the selectivity criteria are tantamount to upfront conditionality and not compatible with Bank initiatives to foster ownership (Bretton Woods Project, 1999). 7 The CDF also seeks to bring NGOs on board into Bank programmes ± prior to Wolfensohn's Presidency, many NGOs tended to see themselves as in opposition to Bank programmes.
Anderson, R.W., C.L Gilbert and A. Powell (1989). `Securitisation and Commodity Contingency in International Lending', American Journal of Agricultural Economics, 71 (Supp.): 523±30 (1991). `Securitising Development Finance: The Role of Partial Guarantees and Commodity Contingency', in T. Priovolos and R. C. Duncan (eds.), Commodity Risk Management and Finance, Oxford: Oxford University Press Bates, R.H. and A. O. Krueger (1993). Political and Economic Interactions in Economic Policy Reform, Oxford: Blackwell Bretton Woods Project (1999). `Bretton Woods Update', London, March Eade, D. (1997). Capacity-Building, an Approach to People-Centred Development, Oxford: Oxfam Eaton, J. and M. Gersowitz (1981). `Debt with Potential Repudiation: Theoretical and Empirical Analysis', Review of Economic Studies, 48: 289±309 Eaton, J., M. Gersowitz and J. E. Stiglitz (1986). `The Pure Theory of Country Risk', European Economic Review, 30: 481±513 Eberstadt, N. and C. M. Lewis (1995). `Privatising the World Bank', The National Interest (Summer): 14±18 Fafchamps, M. (1996). `Sovereign Debt, Structural Adjustment and Conditionality', Journal of Development Economics, 50: 313±35 Gavin, M. and D. Rodrik (1995). `The World Bank in Historical Perspective', American Economic Review, Papers and Proceedings, 85: 329±34 Gilbert, C.L., R. Hopkins, A. Powell and A. Roy (1996). `The World Bank: Its Functions and its Future', ESRC, Global Economic Institutions Working Paper, 15 Hopkins, R., A. Powell, A. Roy and C. L. Gilbert (1997). `The World Bank and Conditionality', Journal of International Development, 9: 507±16 Killick, T. (1998). Aid and the Political Economy of Policy Change, London: Overseas Development Institute
R. Hopkins, A. Powell, A. Roy and C. L. Gilbert
Lessard, D.R. (1986). `The Management of International Trade Risks', Geneva Papers on Risk and Insurance, 11: 255±64 Mosley, P. (1987). `Conditionality as Bargaining Process: Structural Adjustment Lending, 1980±86', Princeton Studies in International Finance, 168 Mosley, P., J. Harrigan and J. Toye (1995). Aid and Power: The World Bank and Policy-Based Lending, 2nd edn., London: Routledge Naim, M. (1994). `The World Bank: Its Role, Governance and Organisational Culture', in Bretton Woods Committee, Looking to the Future, C273-86, Washington, DC Oliver, R.W. (1971). `Early Plans for a World Bank', Princeton Studies in International Finance, 29 (1975). International Economic Co-operation and the World Bank, London: Macmillan Rodrik, D. (1995). `Why is there Multilateral Lending?', in M. Bruno and B. Pleskovic (eds.), Annual World Bank Conference on Economic Development, Washington, DC: World Bank: 167±93 Stern N. F. and Ferreira (1993). `The World Bank as an `Intellectual Actor''', STICERD, Development Economics Paper, 50, London School of Economics Walters, A. (1994). `Do We Need the IMF and the World Bank?', Current Controversies, 10, London: Institute of Economic Affairs Wolfensohn, J. (1999). `A Proposal for a Comprehensive Development Framework ± A Discussion Draft', available at
Conditionality, dependence and coordination: three current debates in aid policy PAUL COLLIER
1 Introduction In this chapter I distinguish between aid transfers and aid relationships. The traditional literature on the economics of aid focused on the resource transfer. Aid relaxed constraints on economic performance, either by increasing savings or by increasing foreign exchange. There was no relationship between donor agencies and recipient governments. Aid, in this analysis, was indistinguishable from a government-owned oil well. I consider three aid debates in which the central issue is the relationship between the donor and the recipient. The ®rst of these is policy-based lending, or `conditionality'. This relationship has been criticised both as intrusive and ineffective (although these two criticisms sit together uncomfortably). The second debate is on `aid dependency'. This criticism is that the aid relationship is intrinsically undermining of national capacities, analogous to the weakening of household capacities in the syndrome of welfare dependency. The third debate is around the suggestion that aid is missing an opportunity for a coordinating relationship. By focusing upon individual nations aid has, it is argued, missed its comparative advantage in the ®nancing of coordinated development at the supra-national level, such as region-wide transport systems. I discuss these three debates in turn.
2 Policy conditionality Although policy conditionality became established only during the 1980s, the idea that international public resources should be used to induce policy reform has a long history. Its origin lies in Fund crisis pro299
grammes. For example, in one of the largest IMF programmes ever undertaken, in 1976, the British government was provided with ®nance on condition that it changed economic policies. There is a continuum from the imposition of conditions on credit expansion in return for an IMF loan at a time of crisis, through to the imposition of conditions on detailed distributional and development policies in return for aid in normal circumstances of growth. However, although policy conditionality started with the former, by the 1990s it had advanced to the latter. In principle this might have succeeded. Indeed, chapter 11 in this volume, by Hopkins et al., assumes that World Bank conditionality is effective, and argues that the resulting complementarity between its development and banking functions provides a rationale for the Bank. In practice conditionality did not succeed. I ®rst discuss why it failed. I then draw a distinction between appropriate and inappropriate circumstances for conditionalities intended to change policy. Finally, I consider a different rationale for conditioning aid allocations on policy ± namely that aid is more effective in some policy environments than in others.
2.1 Why policy conditionality was introduced, and why it failed During the 1980s three factors led donors to introduce policy-based lending. First, donors correctly perceived that in the poor policy environments which had built up during the previous decade, project aid was ineffective. By conditioning aid upon policy change they were attempting to rehabilitate the policy environment. The analytic literature developed formal principal±agent models to characterise this new relationship in which differences between donor and government preferences were central. The decision problem was how the donor could set incentives so as to induce policy change from the government. Secondly, because of a combination of poor policy and negative external shocks, some countries were facing dif®culties servicing their debts. Some creditors were willing to contemplate defensive lending, but this faced a further dif®culty. In some countries, the government administration was unable to cope with the bureaucratic burden of increased project aid. Much project aid could simply not be disbursed. Hence, defensive lending through increased project aid would be ineffective. An important motive for the Special Program of Assistance to Africa was that lending based upon the promise of policy change could disburse large sums at minimal administrative burden.
Conditionality, dependence and coordination
Thirdly, some countries descended into crisis and needed IMF programmes to extricate them. As in all IMF crisis programmes, there was a reasonable presumption that the government had made serious mistakes so that it should be required to make policy changes in return for the ®nance that it was seeking. Since donor ®nance was typically part of the rescue package, the development conditionalities with which donors were concerned could simply be tagged on to this existing medium of policy conditionality. The ®rst two factors inducing the introduction of policy-based lending, the inducement of policy change and defensive lending, were incompatible. If governments did not adhere to their promises of policy change then the donors would be faced with a choice between discontinuing lending and thereby causing loan default, or forgiving the conditions. Since donors valued two outcomes which were incompatible, even had they been able to make this choice in their own best interests, it is unclear what the decision would have been. Quite possibly, the desire to avoid default would have been given more weight that the desire to induce policy change. However, the actual decision structure in the International Financial Institutions (IFIs) and the donor agencies produced a bias in favour of forgiveness. In each particular instance of a breach of conditions enforcement is time-inconsistent: the decision by a donor not to disburse will not directly produce any improvement in government policy, but it will directly lead to default. Enforcement has public good properties: each time a condition is forgiven, expectations of enforcement are reduced and so breaches of conditions become more likely. Usually, the time-consistency problem is presented as a two-agent relationship. However, in the case of donor± government relations, it is in reality a matter of the behaviour of a system. Forgiveness of a government by one agent increases the likelihood of breaches in conditions not only by that government, but also by other governments against other agents. Hence, whether the time-inconsistency problem is overcome depends upon whether each donor agency can internalise the externalities of each particular decision so as to defend the integrity of the system. In practice, much of the power of decision on whether conditions are waived must be decentralised to the country team within the donor agencies, because the country team has an informational advantage over higher managerial levels. This has the consequence that any behavioural change which the decision might induce in other borrowing governments is external to the decision process. Further, because the staf®ng of country teams typically changes every three years, the team has little incentive to take into account any behavioural change which the decision might induce in the government itself. Because
the remaining incentive for the country team is to avoid default, there is an unresolved time-consistency problem. This is compounded because staff have tended to be rewarded for getting loan agreements approved. At the apex of the IFIs are the governing boards, which can take a system-wide view. However, during the 1980s Cold War politics meant that the system-wide view which those delegates with the most votes took re¯ected strategic considerations rather then the effectiveness of conditionality as a means of changing policies. The Cold War supported an aid game in which the leading protagonists competed for allegiance, and in which adept players such as President Nyerere of Tanzania were able to attract large uncritical aid in¯ows from a wide political spectrum. There is considerable evidence for such lending behaviour. For example, the World Bank appears to have ®nanced the same Kenyan agricultural pricing reform ®ve times in 15 years.1 More generally, Svensson (1998) demonstrates that during the 1980s heavy indebtedness was far more important in determining further lending than was the policy environment. Even without conditionality most governments would have gradually been reforming their economic policies. Partly, reform was induced by the simple experience of economic failure. For example, the African countries which were the strongest reformers from 1987 onwards had policy ratings which were much worse than average just prior to reform (Collier and Pattillo, 1999). Partly, reform was induced by the increasing examples of market-assisted growth and the collapse of the socialist model. Partly, as democratisation spread, governments were forced to become more responsive to their populations so that the political costs of economic failure were increasing. Hence, the background to conditionality was of other forces for policy improvement. Given the evident concerns of donors for matters other than policy change, it would not have been unreasonable for governments to regard compliance with conditions as optional. In a context in which some reforms were happening anyway, and there was in any case plenty of noise in the system owing to external shocks, governments could expect to be able to gain forgiveness from donors because they had done some reform, even if not as much as had been envisaged, and because there had been unforeseen circumstances. Because the other forces inducing policy change coincided with conditionality, the era of conditionality was one of considerable policy reform. However, the test of how much conditionality contributed to this process is the country-by-country relationship from aid ¯ows and policy change. Burnside and Dollar (1997) ®nd that there has been no
Conditionality, dependence and coordination
overall relationship from aid ¯ows onto policy change: aid did not succeed in buying reform. Recall that the impetus for conditionality had been the correct perception that projects failed in the poor policy environments prevalent by the beginning of the 1980s. Two phases of aid failure can therefore be distinguished, the pre-conditionality phase in which aid supported projects which failed because of the policy environment, and the conditionality phase, in which aid was targeted on buying policy reform in poor environments, but failed to do so. As a result of these two failures, overall, it is not surprising that aid should have had no signi®cant effect on growth (Burnside and Dollar, 1997).
2.2 The opportunity to rethink conditionality Four factors have radically changed the potential for aid to be effective. Between them, they have created an opportunity for conditionality to be redesigned. First, with the end of the Cold War it has become more feasible for developed country governments to rebase their aid policies on considerations of long-term poverty from considerations of political allegiance. Of course, there is continuing pressure on bilateral donors to favour their own commercial interests, but there is a constituency to oppose such pressure. For example, the current British government has rebased its aid policy on the UN agreed goals, explicitly severing the previous link with commercial interests. Secondly, there have been concerted efforts to reduce indebtedness, currently through the Heavily Indebted Poor Countries (HIPC) initiative. As this proceeds, it will considerably reduce pressures for defensive lending, since it is designed to reduce debt service to manageable levels. Thirdly, the gradual improvement in economic policies in many developing countries has brought some countries to the stage at which the policy environment is now satisfactory (though certainly capable of further improvement). In low-income sub-Saharan Africa, around 25 per cent of the population lives in such environments. The average growth in per capita GDP for this group during the most recent two years for which data are available (1995 and 1996) was 4.2 per cent p.a. Hence, reform has brought in a phase of quite rapid growth in some, though not all countries. Because the reforming countries started from the worst investor risk ratings, private investment has yet to rise to levels which could sustain these growth rates. Growth is currently high owing
to a bounce-back from poor policies. This creates a role for aid in sustaining growth in these newly post-stabilisation economies until private investment rises. Fourthly, as a result of research, we now know much more than we did on aid effectiveness. Burnside and Dollar (1997 and chapter 8 in this volume) show that to be effective in raising growth and reducing infant mortality, ®nancial transfers should be conditioned not upon the change in the policy environment, promised or achieved, but in its level. Below a certain level of policy aid simply fails to achieve growth. The current area of research uncertainty is as to which policies are most important. Burnside and Dollar used a composite of only three aspects of macroeconomic policy: economic openness, ®scal order and the containment of in¯ation. However, they recognised that because many policies are quite highly correlated, this narrow range of macroeconomic policies might itself proxy a wider, but less measurable range. The World Bank has recently constructed standardised and comparable numerical measures for 20 aspects of policy, across the entire spectrum of government for all its client countries. These measures are con®dential, but some research has already been done on their properties. First, they are indeed highly correlated, so that a simple unweighted average of the 20 measures is virtually identical to the ®rst principal component. Secondly, replicating the Burnside and Dollar study on growth performance during the 1990s but replacing their narrow measure of policy with the simple average of the 20 policies yields qualitatively the same results (see Collier and Dollar, 1999).2 Policy and aid interact, so that the better is the policy environment the more effective is aid. In the best policy conditions aid powerfully raises the growth rate. Having been ineffective for so long, the above four changes provide a choice between two aid strategies, either of which could be effective. One option is to build on the opportunity provided by debt forgiveness and the end of the Cold War to make the previous form of conditionality work. This time, breaches would not be forgiven, because the impetus for forgiveness is much reduced, so that the Boards of the IFIs could focus on development performance. However, there are four factors counting against this option. The ®rst is that probably the main reason why conditionality failed was that it under-estimated the importance of the domestic political forces which determine policies in the long-term. There has been a tendency to overestimate donor power over governments. Governments and the `left' has exaggerated donor power in order to raise the spectre of neo-colonialism, and the IFIs have been reluctant to admit that they have considerably less power than such accusations imply. Major economic policies
Conditionality, dependence and coordination
(those which have powerful distributional consequences) are in the long term determined by a domestic political process in which the calculus of aid is often marginal. Recent work on modelling aid bargaining between the donor and the government explicitly allows for a political equilibrium between proreform and anti-reform forces within the society (Coate and Morris, 1996; Adam and O'Connell, 1997). The new analytic approach is to search for donor interventions which irreversibly strengthen the proreform lobby ± for example, by inducing irreversible investment in the export sector. However, while this is analytically correct, its main operational message is surely that sustainable reform in the context in which a government is opposed to it, is radically more dif®cult than temporary reform. A non-democratic government is likely to have a longer time horizon than the country team of an aid agency. Country teams seldom last for more than three or four years, and this limits the scope for credible long-term threats on the part of the donor, although potentially management systems might enable the continuity of aid agency policy to outlast the country team. The second factor counting against the rehabilitation of conditionality is that the record of donor and IFI behaviour has made it much harder for conditionality to be credible. The aid industry lacks any institutions which are capable of credibly binding themselves to new behaviour. For years, donors have complained about non-compliance and made menacing noises about getting tougher. In practice, were donors now to announce that henceforth they would enforce conditions, there would be a phase during which their resolve would be tested by continued non-compliance. A phase of non-compliance which triggered reductions in aid would be costly. The third factor is that the reform agenda has largely moved on from macroeconomic policies to sectoral policies, such as civil service reform and privatisation. These policies are intrinsically more complex, both politically and administratively, and so less suited to timetabled conditions than the earlier generation of reforms. The fourth factor is that using aid to buy policy reform has an opportunity cost in that it precludes both other uses of the resources and other styles of donor±government relationship.
2.3 IFI relations with governments: IMF relations during crisis management compared with development agency relations Recall that one impetus for donor conditionality was the easy link with IMF crisis programmes. However, the IMF is in something of a unique position in its relationship with governments. The IMF is invited in only once a government is in a crisis, and this is usually fairly clearly at least partly of the government's own making. Immediate policy change is inevitable, and it is reasonable for the Fund to treat the government as part of the problem which needs to be overcome. Hence, money for policy change is a sensible stance. The time-consistency problem is considerably less severe in the context of ®nancial crisis. Governments do not choose to enter ®nancial crises, they stumble into them through incompetence, and invariably pay a high political price with or without IMF ®nancial support. The government may at some future date again lapse into crisis, but this will re¯ect not the Machiavellian cunning implied by the time-consistency problem, in which a government rationally chooses to break its promise, but rather the persistence of incompetence. The IMF is not a development ®nance institution, in contrast to the World Bank, the European Union, the UN and the bilateral donors. While development ®nance institutions must necessarily take a view on appropriate development policies, their goals are long-term and their relations with governments are not primarily concerned with crisis management. The scope of pertinent policy is much more wide-ranging. As a result, the scope for the time-consistency problem is much more severe than during ®scal crises. A non-democratic government may quite rationally pursue policies which sacri®ce both growth and poverty reduction for elite interests, and enhance resources available for the elite by reneging on promises. Faced with such a government the true implication of the time-consistency games is that outside the context of crisis there is little which aid can do to induce a government to undertake sustained policy change. In democratic societies there are unlikely to be substantial sustained con¯icts between the objectives of the government and the objectives of the donors because the overriding goal of donors is that governments should be responsive to their own electorates. Donors may occasionally lay claim to ethical superiority over democratic governments ± for example, with respect to the treatment of minorities or the environment ± but such cases are exceptional. Substantial disagreements between the two are then more likely to arise because of differences in information or ideology. In many respects governments are likely to know best because they
Conditionality, dependence and coordination
are closer to the concerns of their citizens, and because they know their economy better than external agencies. In some respects, however, external agencies have an informational advantage. The IFIs are able to recruit better quali®ed staff, and these staff have access to comparative material as to what has worked elsewhere. Since neither party has an absolute overall informational advantage, while both parties have similar overall objectives, the appropriate model for donor±government relations is not that of a game played between an altruistic donor and a recalcitrant government. Rather, it is one of partnership between two agents who gain from cooperation. In such a partnership, threat±promise negotiations are quite dysfunctional since they signal that the donor either has different objectives from the government, or that the donor considers itself to have informational dominance, with nothing to learn from the government. The appropriate stance for donors in democratic societies is thus to convince and to listen rather than to cajole. This will obviously not always be successful. There is a market place in ideas, and sometimes dysfunctional populist economic policies will gain the ear of democratic governments and of democratic societies. Donors should not, in such circumstances, lapse into the mode of offers and threats unless the economy collapses into crisis. Rather, they should improve their knowledgedissemination and bide their time. This, of course, cannot be the approach of the IMF during the onset of a crisis, but most of the time developing countries are not in crisis. By abandoning the notion of aid as a `reward' for policy improvement, donors move to a model of partnership. However, with those governments which adopt really poor policies, partnerships are not bene®cial. Engagement with those governments and with their societies in the battle of ideas is the means by which donors can best hope to in¯uence policy. Where governments have reasonable policies donors need to engage both ®nancially and through knowledge. The presumption must be that there will be continuous differences in opinion, re¯ecting the fact that each party has very different knowledge, with neither party having full knowledge. A relationship of trust, in the sense that each side is con®dent that the goals are shared, is more likely to be conducive to mutual learning. The threat±promise mode of negotiation is destructive of trust because it presupposes that the donor believes that it has difference preferences or globally superior knowledge. I am thus suggesting that donor conditionality should, to be effective, take a very different form from IMF crisis management conditionality. In the latter, the Fund can reasonably be supposed to know better than the government and, government incompetence having been demonstrated,
can reasonably coerce the government until the economy is stabilised with as little hardship to the poor as is possible. In the former, donors should condition their ®nancial resource transfers not upon policy change but upon policy levels. Their engagement in policy debate should, however, be at its most vigorous where policy is worst, so that aid and knowledge can be sequenced.
2.4 The crisis-management role of the donors The Fund cannot stay out of economies in crisis ± that is, after all, its core rationale. By contrast, the development ®nance institutions should ®nd crisis environments unattractive for ®nancial resources intended to secure growth and poverty alleviation through growth. There are, however, two rationales for donor ®nancing of crisis management. First, there will sometimes be a case for emergency assistance which directly alleviates poverty by mitigating the decline in the consumption of the poor even though it does not achieve sustainable growth. This is closely analogous to humanitarian assistance in response to natural disasters. Secondly, the success of a Fund stabilisation programme may be dependent upon larger resources than the Fund has at its disposal. This is not inevitable: beyond a certain level of resources additional money may simply reduce the need for the government to take corrective action. The appropriate aggregate level of funding at which the chances of stabilisation are maximised is a complicated judgement. Once the Fund has determined the right level of ®nancing, if the amount is beyond its resources, it should attempt to broker assistance. Such assistance might well come from the same institutions which are involved in development ®nancing, just as humanitarian emergency assistance will often come from aid agencies. However, just as humanitarian assistance has to be negotiated in a different mode from development assistance, so stabilisation funding may need to be treated as a distinct entity from development assistance.
2.5 The post-stabilisation role of the Fund Until recently, the proposed different specialisations of the Bank and the Fund would have been unproblematic since Fund engagement was con-
Conditionality, dependence and coordination
®ned to crisis situations. However, recently, the Fund has extended its involvement to post-stabilisation environments ± for example, Uganda, which came out of economic crisis in early 1992, still has an ESAF programme in 1998. The Fund can play a very useful role in such poststabilisation environments, but it is not the same as its role in crisis management. Instead of being there to cajole a recalcitrant government, the main function of the Fund is to assess government macroeconomic policies, freely chosen and, where appropriate, reassure the domestic and international investment community that policies are sound, thus accelerating the recovery of reputation and investment. The role of the Fund thus evolves from that of coercing policy change to accreditation of policy levels. The role of accreditation is particularly important in post-stabilisation low-income economies. Investor con®dence is often at a low level. Policy reform can achieve rapid growth even with low levels of investment for a few years, but in order for growth to be maintained investment must recover. The government needs to establish a reputation as a competent economic manager, demonstrating that it has learnt its lesson from crisis. In order to establish its reputation, a government needs gradually to be given the overt power to initiate policy reform, while maintaining macroeconomic order. For this, it needs to be released from the coercion mode in which its good actions are seen as being the result of force majeur. At some point, the Fund might thus usefully signal that its role has changed from that of crisis manager to that of post-stabilisation evaluator. Obviously, the passage from crisis management to post-stabilisation is a continuum. However, in many circumstances it is useful to convert continua into discrete phenomena even though the dividing line will be arbitrary. The advantage gained is in reducing the costs of information. For example, the European Union converted the continuum of a ®scal de®cit measured as a proportion of GDP to the target of 3 per cent. This created an identi®able hurdle which could thereby be readily monitored by electorates: the target solved the coordination problem as to what level of de®cit was unsatisfactory. Analogously, the Fund could usefully create a hurdle between crisis and post-stabilisation environments. The passage to post-stabilisation would signal to investors that the environment should be taken seriously, and to donors that stabilisation-mode funding should taper out, while development assistance funding should taper in.
2.6 Tapering-in of development assistance Assuming that development aid should avoid crisis economies, and those economies in which government policies are so poor as to be heading for crisis, at what stage in policy reform post-crisis should donors come in? Suppose, for the moment, that whether reforms persist is independent of aid ¯ows. In this case the calculation for donors is simple. The Burnside and Dollar results show that there are diminishing returns to aid, as might be expected, and that for a given level of aid returns are higher the better is macroeconomic policy. A reasonable objective for the donor would be to have as high a return on aid money in terms of poverty reduction as possible, and for this the impact of the marginal dollar would need to be equated across aid-receiving countries. The level of aid relative to GDP would thus rise as the policy environment improved. Aid would thus gradually taper in with reform. Donor resources would be provided on the basis of partnership described above, and so would not be implicated in Fund conditionality. Donor resources would start to ¯ow only when the need for Fund conditionality ceased as the economy emerged from crisis. Figure 12.1 shows the relationship between aid and Figure 12.1 The relationship between aid and policy for the median developing country Source: Collier and Dollar (1999: ®gure 3).
Conditionality, dependence and coordination
policy for the median developing country during the 1990s. The ef®cient `tapering in' relationship keeps the marginal impact of aid on growth constant: hence as policy improves, more aid can be productively absorbed. The actual relationship, estimated from global aid ¯ows, is very different. Aid typically comes in too early in the reform process to be used effectively, and is then withdrawn, or tapered out, over precisely the range of policy in which it is effective. Now suppose that aid at the right time increases the chance that reforms will be maintained. At present there is little formal research on this question. However, an important recent discovery is that in satisfactory macroeconomic policy environments aid leads in private investment: an extra dollar of aid increases private investment by $1.90 (World Bank, 1998). A potential point of fragility in reform programmes is if growth rates decelerate as the initial direct contribution of reform to raising economic activity fades before private investment has risen suf®ciently to maintain growth at this level. In the early post-stabilisation years private investment is too low because of high risk. Once private investment has risen to levels at which rapid growth is sustainable the investment is itself liable to be sustained. In the process of investing, agents acquire further information which enables them to reassess risk. Hence, just as risk discourages investment, so investment would improve the risk ratings. The World Bank now emphasises that it has a dual function both as a provider of ®nance and as a provider of knowledge. In the case of newly stabilised economies it has a considerable informational advantage over private agents, and so has a rationale for using ®nance to raise private investment to rates which will become self-sustaining but which would otherwise not take place prior to a growth deceleration. To give orders of magnitude, in the African good policy group investment is currently only 18.2 per cent of GDP. To sustain rapid growth it needs to be increased by around 10 per cent of GDP. Public investment in Africa is similar as a share of GDP to that in other continents, so the task is to raise private investment by this amount. Using the implied coef®cient from Burnside and Dollar (1997), this would require an increase in aid of around 5.3 per cent of GDP. This would be temporary, until investor con®dence recovered, endogenously. Because aid both raises growth and private investment, there are two relationships which determine its trajectory. The long-term growth relationship links growth to the share of aid in GDP, subject to diminishing returns. Because of diminishing returns, ef®cient use of aid would imply a fairly constant share of aid in GDP, subject only to tapering in as policy improved, and tapering out as income rose. The short-term growth rela-
tionship links investment to aid. In economies in which investment starts well below target levels for sustainable rapid growth, aid would need temporarily to be higher than long-term levels. Hence, aid would taper in much more rapidly than implied by the long-term growth relationship, being reduced back to its long-term path once investment had reached its target.
3 `Aid-dependency' While conditionality sought to use aid transfers as the basis for a relationship which would augment growth over and above the direct effect of the resource transfer, so that conditional aid would be more valuable than an oil well, the aid-dependency argument is that the aid relationship is dysfunctional, so that aid is less valuable than an oil well, and indeed may have overall negative effects. The aid-dependency argument is partly institutional, partly microeconomic and partly macroeconomic. At the institutional level the critique mainly focuses on project aid rather than policy lending. It is that the activities of government become swamped with the requirements of negotiating and administering donordetermined projects. Each donor has its own particular negotiating and reporting requirements, and since much bilateral aid is tied to the use of the consulting and contracting services of the donor country, even within a sector there will be a multiplicity of project designs and makes of equipment. Hence, precisely where the local civil service is in any case too weak to impose its own agenda on donors, it will acquire a portfolio of projects which is so complex that it would challenge even a strong bureaucracy (Kanbur, 1998). At the microeconomic level, the critique focuses on the negative incentives of transfers analogous to welfare dependency at the level of the household. The argument is partly that crises force change and that aid, by averting crisis, enables poor policies to be maintained. Note that this runs counter to the rationale for conditionality, which presupposed that by being linked to policy change, aid would increase the incentives for good policies. The two opposing views can be reconciled analytically by thinking of them as income and substitution effects. The notion that necessity forces reform is that the income elasticity of demand for reform is negative: a reduction in income increases reform. Conditional aid raises income, and so its income effect is to reduce the incentive for reform, but if the conditionality is credible it also has a relative price effect which increases the payoff to reform, so that there
Conditionality, dependence and coordination
is a substitution effect inducing policy change. Which of these two effects predominates can be settled only empirically. A related argument is that because aid reduces tax `effort' it weakens domestic political pressure on government. Taxation has historically been important both in achieving democratic government (`no taxation without representation') and in increasing the incentive of electorates to hold governments to account. While even the government is spending aid money, the electorate is less concerned about corruption and a failure to deliver public services. At the macroeconomic level the argument is partly that aid appreciates the exchange rate and so implicitly taxes the export sector. Since the export sector is an important engine of growth, aid could thereby reduce growth. A further argument is that aid is volatile: donors are ®ckle, with aid ¯ows interrupted for political reasons, so that an aid-dependent country is subject to more shocks, especially to government resources. Since governments are particularly bad at managing their own income volatility, aid shocks can destabilise the economy.
3.1 At what point does a country start to drown in aid? Aid, like most other things, is subject to diminishing returns. Hence, the hypothesis of the aid-dependency arguments is not that `aid is harmful' but rather, that `big aid is harmful', with the implicit question hanging, `at what point does aid switch from being useful to being harmful?' Of course, part of the aid-dependency argument is that this depends upon how aid is delivered. As Kanbur argues, aid delivered as budget support has many fewer institutional problems than aid delivered as projects. Since to date project aid has been much more important than budget support, there is considerable potential for defusing the institutional critique by changing the composition of aid, instead of reducing its overall volume. Collier and Dollar (1999) estimate diminishing returns to aid on a panel of observations for 1970±93, introducing aid/GDP as a quadratic. They ®nd that, while aid is indeed subject to signi®cant diminishing returns, the point at which returns reach zero depends upon the policy environment. In poor-policy environments aid soon becomes ineffective. In good-policy environments, however, quite large amounts of aid can be productively absorbed. Typically, with good policies, a country starts to `drown' in aid only when the share of net aid in¯ows to GDP exceeds around 12 per cent, which is a very high ®gure.
Hence, in good-policy environments `big aid' is effective. Further, `big aid' interacts favourably with policy improvement. A new study (World Bank, 1998) compares a country with an average amount of aid with one which has double the average. In both cases it investigates the effect of a given improvement in policy. The country with `big aid' gets a 50 per cent larger increase in its growth rate from the policy reform than the country with average aid. Hence, on this evidence, if anything `big aid' appears to help a government to manage change rather than distract it. Recall, however, that Burnside and Dollar (1997) ®nd no link from aid ¯ows to policy reform itself: aid does not induce reform, although it does increase the payoff to reform. Thus, within the range of aid ¯ows which is pertinent, the evidence suggests that, subject to a satisfactory policy environment, the more aid which is provided the faster will an economy grow.
3.2 Is aid-dependency like welfare-dependency? If aid really had the sort of disincentive effects on poor countries that welfare payments have on poor households, aid should reduce growth irrespective of the policy environment. Indeed, we might expect that the better the policy environment, and so the higher the potential growth rate in the absence of aid, the more damage aid would do. However, the analogy with welfare-dependency is not just empirically questionable, it is more fundamentally mistaken. First, the scale of aid relative to income is tiny by comparison with welfare payments. The `poverty trap' of welfare-recipient households commonly involves households for whom welfare payments constitute a large majority of total income and who thereby face implicit marginal tax rates of 80 per cent or more. By contrast, gross aid ¯ows to Africa peaked at around 12 per cent and implicit marginal tax rates are correspondingly radically lower. Alesina and Dollar (1998) analyse the patterns of donor aid allocation. The donors who most sharply reduce aid in response to rising incomes are the Nordic group. Their aid ¯ows start to fall once per capita incomes rise above $600 and are largely eliminated once incomes rise above $1,600. However, since the loss of aid involved in this policy is only around $3 per capita, the marginal tax rate implied by Nordic policy is well under 1 per cent. Similarly, USAID tapers out its, historically, much larger aid budget but it does so much more gradually. The implicit tax rate imposed by USAID is less than one-half of 1 per cent. Other donors currently are less responsive to changes in income.
Conditionality, dependence and coordination
Hence, the aggregate marginal tax rate implied by donor reductions in aid as income rises is probably in single digits. Any disincentive effect at the aggregate level is therefore trivial. Further, incentive effects can operate only at the level of individual decision-takers, rather than the national aggregate. Suppose for the moment that, implausibly, it was indeed the case that for each extra dollar of (say) Zambian national income, aid declined by a dollar. At the aggregate level there would therefore be a 100 per cent marginal tax rate and so no incentive whatsoever to work. Nevertheless, this would not create a poverty trap for Zambians. This is because Zambians do not decide their income collectively. Each individual Zambian household, if it earned an extra dollar, would lose only its in®nitesimal share (less than a millionth) of the dollar of aid which Zambia in aggregate would lose. The loss of aid would therefore take the form of negative externalities to individual household decisions. In effect, Zambians would be locked in a `work trap': households would work even though in aggregate it made them no better off. The one exception to the absence of an incentive effect of aid is with respect to the behaviour of governments. Directly, aid accrues to the government and so it indeed faces an incentive problem vis-aÁ-vis taxation. In the absence of aid a government has to balance the political unpopularity of taxation against the political popularity of public expenditure. At the margin, a dollar of tax revenue has as many political costs as a dollar of expenditure has political bene®ts. The receipt of aid by the government does not change this fundamental balancing requirement of political economy, but it necessarily changes the levels of taxation and expenditure at which balance is achieved. Figure 12.2 depicts the rising political costs of taxation by the T-T schedule, and the diminishing political bene®ts of public expenditure along the E-E schedule. The intersection of the two schedules at E0 is the equilibrium without aid. The with-aid equilibrium, Ea, involves both higher expenditure and lower taxation than without aid. If the political process broadly encapsulates the economic costs and bene®ts of taxation and expenditure, then this response is optimal. The ®rst inference from this is that aid will indeed induce governments to have lower taxation than without aid. In effect, the population is choosing that some of the bene®ts of aid accrue to households most ef®ciently by passing the money on to them to spend, by means of reducing taxation, while some of the bene®ts of aid accrue to households most ef®ciently through increased public expenditure. It is hard to see this as an incentive problem since it is the solution to a household welfare-maximising problem which is identical to that which faces the donors. A priori, donors have no particular reason to prefer aid to accrue directly or indirectly to
Figure 12.2 The ®scal effects of aid
households. However, the reduction in taxation has favourable incentive effects at the household level. The marginal costs of taxation are often quite high in developing countries because there are few ef®cient tax handles. Hence, reduced tax effort can free households and ®rms from powerful disincentive effects of taxation. Thus, precisely counter to the welfare-dependency analogy, aid actually improves the incentive environment at the level of the individual agent. A simple extension of this argument is the more familiar `debt-overhang' effect. Debt, which is simply future negative aid, discourages investment because it constitutes a future tax liability. Hence, it is increased debt rather than increased aid which gives rise to an incentive problem: the counterparts to welfare households trapped into idleness are ®rms trapped into disinvestment.
3.3 Is aid a source of ®scal instability? I now turn to the macroeconomic arguments on aid dependence. The effect of aid on the appreciation of the real exchange rate, though obviously correct, does not imply anything about the net effect of aid on growth. Rather it points out one negative channel. The empirical resolution has already been discussed above: in good policy environments
Conditionality, dependence and coordination
the net effect of aid is to raise the growth rate until very high levels of aid are reached. The second macroeconomic effect of aid is more contentious, namely that it destabilises the budget. This is indeed the reason given by the IMF for treating aid receipts as an exceptional ®nancing item to meet a budget de®cit, rather than as a core component of non-tax receipts. The ®scal de®cit is reported excluding aid because aid cannot be relied upon. There are two main reasons why aid receipts might be unreliable. One is that donors may use aid to advance a political agenda driven by the political concerns of their domestic electorates. Secondly, donor procedures for disbursement may be so cumbersome that, even when funds are committed, there may be long and unpredictable lags before governments are able to utilise them. There are good grounds for both of these concerns. For example, the aid cutoff to Pakistan when it matched India's testing of nuclear weapons constitutes a major ®scal shock which could not reasonably have been anticipated in budgetary planning. Similarly, both CoÃte d'Ivoire and Uganda received entitlements to Stabex funds from the EU as a consequence of the fall in world coffee prices in 1989. However, CoÃte d'Ivoire was able to gain access to these funds in a reasonably timely fashion, whereas disbursement to Uganda was so delayed that much of the money was received during the boom in world coffee prices ®ve years later. However, while such stories demonstrate that there is some basis for concern, similar stories could be told about the unreliability of tax revenues. African governments are highly dependent upon trade taxes for their revenue, typically accounting for half of overall receipts. Yet trade taxes are dependent upon the capacity of the economy to export and import. African exports are highly concentrated in a narrow range of commodities the prices of which are volatile. Hence, exports are much more volatile than GDP as a whole. The capacity to import is dependent partly upon exports, and partly upon aid. Hence, a priori, it is not clear that aid would be signi®cantly less reliable than government revenue. Even if aid is less reliable than government revenue, it might nevertheless reduce the overall unreliability of the resource ¯ow to the government if it moves inversely with revenue. That is, as with any portfolio, an important consideration is not just the variance of each component, but whether the risks are co-variant. Hence, in measuring the reliability of aid both its variance and its covariance must be taken into account. I measure the relative reliability of aid and government revenue for 36 African countries which are IDA recipients over the period 1970±95. I standardise by measuring both aid and revenue year-by-year in 1995 US$
per capita. The weighted average for these countries over the period was that aid amounted to $72.4 while government revenue was $279.7. In assessing the reliability of aid and revenue we need some measure of unpredictability. In principle, unpredictability is very different from variability: for example, if either aid or revenue has a regular cycle it could be highly variable yet highly predictable. One approach would therefore be to model both aid and revenue as time series, determining how well they could be predicted econometrically, and taking the regression errors as the measure of `unreliability'. However, this is not the approach followed here. In practice, both governments and the IFI teams that assist them do not make such predictions, often for the good reason that the results would lack robustness. Hence, I rely upon a much more straightforward measure of unpredictability namely, variance. The measure of variance which scales for the mean level of receipts is the coef®cient of variation (standard deviation/mean). I thus calculate the coef®cient of variation of aid receipts and revenue both for each country and for the weighted average of all 36 countries. The results are reported in table 12.1. The ®nal column of the table reports the normalised covariance of aid and revenue (covariance/the product of the means). The results are, of course, country-speci®c. However, in assessing the future riskiness of aid policy-makers may well regard the information from the entire group of African countries a better guide than the historical experience of just their own country. After all, the move from a single country to the full sample represents the move from 25 observations to 900. The best guide to the full sample is the weighted average. The coef®cient of variation of aid is lower than that of revenue. Aid is more reliable than revenue, not less reliable as the aid-dependency school believes. Secondly, the normalised covariance of aid and revenue is negative (although not signi®cantly different from zero). This implies that there is a further bene®t from aid: it acts as a buffer to revenue shocks ± if anything, tending to increase when revenue is low. More importantly, being essentially uncorrelated with revenue, dependence upon aid provides bene®ts of diversi®cation and so is stabilising. Hence, on the aggregate evidence, a budget with a large component of aid would be more reliable than one with a small component of aid, because the aid component is more certain than revenue, because it tends to offset revenue shocks, and primarily because government funding is more diversi®ed. In some countries the contribution of aid to the overall stability of government resources is striking. For example, in Uganda the coef®cient of variation of aid is only one-tenth that of revenue. In such a case, if reliability was the criterion for inclusion in the budget the ®scal de®cit
Conditionality, dependence and coordination
Table 12.1 The coef®cient of variation of revenue and aid in 36 African countries, 1970±1995 Aid ($ per capita)
Revenue ($ per capita)
Covariance/ Product of means
Coeff. of variation
Coeff. of variation
Burundi Benin Burkina Faso Botswana Cameroon Congo Comoros Ethiopia Gabon Ghana Guinea Gambia Guinea-Bissau Kenya Liberia Lesotho Madagascar Mali Mauritania Mauritius Malawi Niger Nigeria Rwanda Sudan Senegal Somalia Swaziland Seychelles Chad Togo Tanzania Uganda Zaire Zambia Zimbabwe
42.7 44.7 44.6 150.7 47.0 82.9 167.6 18.7 169.9 35.2 77.0 104.2 123.6 39.3 67.3 76.7 37.2 60.1 265.9 73.1 39.8 65.1 2.6 48.2 43.1 75.9 50.2 103.2 399.3 49.2 67.7 46.0 18.0 19.9 82.5 37.2
0.29 0.18 0.22 0.31 0.29 0.33 0.05 0.32 0.30 0.51 0.12 0.44 0.23 0.34 0.35 0.26 0.28 0.21 0.34 0.28 0.39 0.23 0.65 0.22 0.52 0.31 0.43 0.38 0.30 0.21 0.30 0.46 0.07 0.38 0.57 0.59
39.8 86.1 36.4 853.1 198.2 324.9 69.5 33.3 2493.1 65.8 80.6 86.4 0.4 105.4 175.5 121.1 66.5 42.3 193.7 478.1 50.6 88.8 216.6 39.4 107.4 152.0 38.9 358.2 2718.7 28.3 158.2 74.7 19.3 53.5 206.4 273.1
0.27 0.08 0.21 0.64 0.34 0.65 0.18 0.31 0.39 0.37 0.10 0.34 0.20 0.29 0.23 0.39 0.85 0.21 0.08 0.33 0.17 0.28 0.51 0.30 0.20 0.24 0.23 0.33 0.27 0.40 0.34 0.20 0.70 0.68 0.69 0.25
0.0156 0.0101 0.0330 0.0729 0:0515 0.1173 0:0024 0:0279 0:0258 0:0124 0:0080 0.0288 0.0159 0.0083 0.0028 0.0624 0:0417 0.0102 0.0199 0.0040 0.0214 0.0121 0.0333 0.0294 0.0199 0.0290 0:0314 0.0806 0:0562 0:0370 0.0231 0.0465 0.0011 0.1249 0:2150 0:0390
Aid with revenue
should be reported excluding revenue, rather than excluding aid. More generally, this particular proposition of aid-dependency has little empirical basis.
4 The aid relationship as an instrument for regional coordination Aid has some potential to increase the incentives for governments to coordinate their behaviour. This is at its most important in regions which are Balkanised into many small states. The region most characterised by this is Sub-Saharan Africa, where a population less than half that of India is divided up into more than 40 countries. This has two important consequences. First, the average economic size of a country is very small. As a result there are potentially large-scale economies from market integration with neighbours. Similarly, some government projects are subject to problems of small scale, and so would gain from being amalgamated at the regional level. Secondly, the average geographic size of a country is small. One consequence is that an unusually high proportion of a country's population is proximate to neighbouring countries, so that both populations and diseases can readily move across borders. A crisis in one country becomes a refugee problem in its neighbour, and an epidemic in one country becomes an epidemic in its neighbour. A second consequence is that many countries are landlocked. As a result the provision of the transport infrastructure on which the country depends for its link to the sea is determined by its coastal neighbour, which also has the power to interrupt transport links for political reasons (for example, in 1977 Kenya cut off oil supplies to Uganda). These two considerations of small economic size and small geographic size suggest that there would be atypically high returns from regional cooperation. Presumably because of this insight, African governments have a long history of attempts at such cooperation, mainly for trade, on which there are around 30 agreements, but also for transport such as East African Railways and East African Airlines, and for currency, notably the Franc Zone. However, few of these attempts have yielded sustained success. Partly, this is because regional trade agreements have encountered the problem that, in the presence of high external trade barriers, preferential agreements produce powerful transfers, with the less industrialised countries subsidising the industries of the more industrialised. Partly, it is because, given this history of failed cooperation in
Conditionality, dependence and coordination
trade agreements, the same expectations carry over into other areas where cooperation would be more fruitful. In effect, African governments are in a `low-trust' equilibrium in their relations with each other, expecting that agreements will be broken, and so having little incentive not to break them or to free-ride. Partly, it is because (until the emergence of South Africa) Africa has lacked any obvious countries which know that they are too important to free-ride, and therefore take on a leadership role. There are several ways out from such a lack of cooperation. One is for a lead country within the continent to act as the catalyst and coordinator. In Southern Africa this role is now being played by South Africa, and possibly a democratic Nigeria may assume such a role in West Africa. A second approach is to reduce coordination problems by reducing the number of cooperators: bilateral cooperation is usually easier than cooperation among large groups. However, in some of the important areas for cooperation, notably relations between landlocked countries and their coastal neighbour, bilateral bargaining leaves the negotiating range too wide. The landlocked country has an interest in the coastal government providing a good road. At one extreme, the landlocked economy simply free-rides on the transport investment of the coastal economy. At the other extreme the coastal economy extracts from the landlocked economy the full economic rent associated with international trade, levying road tolls at the revenue-maximising rate. In between, the coastal economy may under-invest in the road, while the landlocked economy may be unwilling to ®nance the road because it faces its own time-consistency problem: once the road has been ®nanced by the landlocked economy, the coastal economy may still extract the full rent for using it. This gives a potential dual role for the international development agencies as catalysts of cooperation and as enhancing the incentives for cooperation. The agencies can provide the leadership role which is missing when several similar countries face a free-rider problem. They can also increase the incentives for cooperation by providing aid ®nancing for regional endeavours. For example, donor agencies could ®nance coastal road links conditional upon certain rules of operation of the road, protecting both the landlocked and the coastal economy from a time-consistency problem. Such a use of aid would depend in part upon it having an effective enforcement effect. This is to an extent analogous to policy conditionality where aid has evidently failed to have such an effect. However, conditionality to enforce regional agreements would be on ®rmer ground. It would have greater moral authority, which could be sanctioned by international consortia of developing countries such as the
UN or the OAU. It would focus on a narrow and precisely monitorable aspect of behaviour.
5 Conclusion In this chapter I have considered three aspects of the aid relationship: conditionality, aid-dependence and inter-government coordination. The ®rst of these has been the predominant relationship of the past decade, while the second has been the predominant criticism. I have argued for a new basis to the aid relationship. Outside the context of crisis, the relationship between governments and aid providers is more productively viewed as a partnership than as a principal±agent problem. I have shown that a partnership between a government with satisfactory policies and `big' aid providers would signi®cantly raise growth, and so would not be subject to the dependency critique. I have suggested that there is also a role for aid as the catalyst and enforcer of regional cooperation. I have attempted to draw a distinction between crisis lending and development assistance. To date, the mode of using aid directly to increase the incentive for policy change, which is appropriate for crisis management, has also been adopted for development assistance. This may well be because the development assistance agencies got into policy-based lending at a time of crisis in the early 1980s. The development assistance agencies came in on Fund stabilisation programmes, but added a development policy agenda. Hence, conditionality spread across a wide array of policies which have little bearing on crisis management. Not only were development assistance institutions functioning in the same mode as the crisis management institution, but also the mode continued postcrisis. As argued by the external evaluation of Fund operations in low income countries, (IMF, 1998) the Fund evolved into a quasi-development assistance agency. It negotiated development policy in post-stabilisation environments across a range outside its traditional area of expertise, but in its traditional mode of relations with governments. Its power in such an environment derived not from its own lending, which was typically only a small proportion of international public transfers, but from the cross-conditionality clauses which the World Bank and the other development assistance agencies kept in their own agreements with the government. Without Fund approval, there would be no aid. The mentality of crisis management has contaminated the much wider process of development assistance in two respects. First, it has contaminated the relationship: as a result, development assistance agencies have
Conditionality, dependence and coordination
been drawn into the attempt to use their resources as an incentive for policy change. Outside the context of crisis it has proved largely ineffective in achieving generalised policy reform, and it precludes the trustbased partnership which would be more appropriate for development assistance agencies. Secondly, it has contaminated the resource ¯ows. The ®nancing of crises is quite properly temporary: stabilisation ®nance should indeed taper out as stabilisation succeeds. By contrast, development assistance should, to be ef®cient in reducing poverty on a sustainable basis, taper in as stabilisation succeeds. To date, as demonstrated in ®gure 12.1, donor ¯ows have followed the short trajectory appropriate for crises, rather than the path appropriate for growth and development. In the post-stabilisation environments which will increasingly be the norm in the developing world, the range of potential policy change is necessarily wide and the mode of relationship is more effective if it is based on partnership rather than coercion. The World Bank has reassessed its role in the development process and has determined a new departure, termed the `Comprehensive Development Framework' or CDF (World Bank, 1999). The CDF is a major attempt to change lending practices to come to terms with these realities and its approach will be piloted in a sample of countries during 1999. It provides for a comprehensive review of the development challenges facing a society, and so explicitly accepts that the range of potential policy change is extremely wide. However, its main innovation is its emphasis upon partnership, both between donors and the government, and between the government and civil society. The CDF is thus a clear departure from the narrow and coercive conditionality appropriate for crisis management and re-establishes the distinctive role of the Bank as a development agency. NOTES
The ®ndings, interpretations and conclusions expressed in this chapter are entirely those of the authors. They do not necessarily represent the views of the World Bank, its Executive Directors or the countries they represent. 1 See Collier (1997). 2 Although the ®rst principal component is close to a weighted average, this does not necessarily imply that this average is necessarily the best policy measure in explaining growth. In Collier and Dollar (1999) we also use a simple average of the 10 policy measures which are each individually signi®cant in the growth process. However, the correlation between this measure and the simple average of the 20 policies is 0.97.
Adam, C. and S. O'Connell (1997). `Aid, Taxation and Development: Analytical Perspectives on Aid Effectiveness in Sub-Saharan Africa', Centre for the Study of African Economies, University of Oxford, mimeo Alesina, A. and D. Dollar (1998). `Who Gives Aid to Whom and Why?', Washington, DC: World Bank, mimeo Burnside, C. and D. Dollar (1997). `Aid, Policies and Growth', Policy Research Working Paper, 1777, Washington, DC: World Bank Coate, S. and S. Morris (1996). `Policy Conditionality', Department of Economics, University of Pennsylvania, mimeo Collier, P. (1997). `The Failure of Conditionality', in C. Gwin and J.M. Nelson (eds.), Perspectives on Aid and Development, Washington, DC: Overseas Development Council Collier, P. and D. Dollar (1999). `Aid Allocation and Poverty Reduction', Policy Research Working Paper, 2041, Washington, DC: World Bank Collier, P. and C. Pattillo (1999). Investment and Risk in Africa, London: Macmillan International Monetary Fund (1998). External Evaluation of the ESAF: Report by a Group of Independent Experts, Washington, DC: IMF Kanbur, R. (1998). `Aid, Conditionality and Debt in Africa', Cornell University, mimeo Svensson, J. (1998). `Aid Tournaments', Development Research Group, Washington, DC: World Bank, mimeo World Bank (1998). Assessing Aid: What Works, What Doesn't and Why', Policy Research Report, Development Research Group, Washington, DC: World Bank (1999). `A Proposal for a Comprehensive Development Framework', Discussion Draft
Abbot, K. 135 accreditation 309 Adam, C. 305 Addison, T. 182±3 adjustment see structural adjustment advice 28, 115, 206 as public good 76 see also knowledge Africa 266, 278, 317 funding level 170±1, 314 good policy group 311 liberalisation 102, 269 Mid-East 241 OED rating 173, 241 policy reform 302 poverty 186 rapid growth 303±4 regional cooperation 320±1 SIP 273±6 African Economic Research Consortium 121 Agranoff, R. 124 agricultural sector 176, 177±8, 187, 266, 273±6 Ahuja, V. 186 aid allocation 224±5 bargaining 305 big 314, 322 coverage 291 and debt-servicing 203 diminishing returns to 310, 313±14
direct and indirect effects 213±15 and economic performance 61±2 effectiveness 26±9, 204±5, 210, 214, 217±19, 267±8 and exchange rates 313, 316 ®scal effects 200±1 and growth 210, 224, 311±12 income and distribution effects 312±13 increase in capital 212, 213±14 and infant mortality 219±23, 224 negative incentives 312±13, 314±16 and policy regime 62, 63, 76±7, 210, 214, 217±19, 221±2 and private investment 311±12 quantity 216±17 and recipient's budget 197, 200, 201 and regional coordination 320±2 tapering-in 310±12 targetting 187, 224, 291, 294±5 transfers and relationships 299 volatility 313, 317±19 see also fungibility; lending; of®cial development assistance (ODA); public expenditure reform loan (PERL) aid-dependency 312±20 AIDS 122 Alba, P. 66 Albania 133 Alesina, A. 224, 314
Alverch, H. 124 Anand, S. 91±2 Anderson, R. W. 285 Annual Review of Development Effectiveness 268 Arabia 121 Arbenz, Jacobo 146 Argentina 69, 117 Ascher, W. 134, 136, 140 Asher, R. E. 13, 15 Asia 94, 159, 167, 170, 171, 173, 186, 241 Asian crisis 68, 70, 72, 102, 268 Asian Development Bank (ADB) 5 Asian Miracle 47 Assessing Aid 128, 268 asset prices 181 Austria 133 Ayres, R. I. 16, 134 bailout options 68 Baker, James 141 Balassa, B. 116 Bangladesh 118 Bank for International Settlements (BIS) 148 banking system 66±7 Barro, R. 233±4 Basu, S. 127 Bates, R. H. 25, 283 Bauer, P. T. 216 Belarus 133 Belgium 133 Bird, K. 134 black market 234±44 Boone, P. 200 Bosnia 138, 139, 147 Botswana 216, 224 Bourguignon, F. 93, 164, 181, 183, 185 Boutros-Ghali, Boutros 148 Brady bonds 16 Brady, N. 16, 141 Brady Plan 16, 141 Braithwaite, J. 52
Branson, W. 169±74, 178, 181, 184±6, 190±2 Brazil 89, 93, 159, 167, 169, 186 Bretton Woods 39, 42, 56, 91, 282 Brown, L. D. 133, 148±9 Bruno, M. 161 Buckley, R. 268 Buira, A. 133 Burma 217 Burnside, C. 8, 27, 32, 61±3, 174, 179, 188, 210±26, 288, 302±4, 311, 314 capacity underutilisation 231 capital increased by aid 212, 213±14 long-term accumulation 182 market imperfections 42, 47, 53 private ¯ows 42 see also investment capital gains and losses 181 Caribbean 170, 173, 241 cash budgets 275 Cashel-Cordo, P. 200±1 Central America 146 Central Asia 117 Chen, S. 92, 118, 180 Chile 117 China 117, 118 Chiwele, D. K. 273, 276 Chong, A. 93 civil liberties 228±9 cross-country indicators 248±53 and ERR 250±1 see also institutions civil strife 251±3 Clark, I. 72 Coate, S. 305 Cold War 146±7, 302, 303, 304 Collier, P. 8, 27, 30, 32, 63, 128, 217, 224, 289, 299±323 Colombia 118 colonialism 3±4 Commission on Dams 53 competitive pluralism 5±6
Index Comprehensive Development Framework (CDF) 6, 29±30, 40±1, 52, 77, 102, 143, 207, 293±5, 323 computable general equilibrium (CGE) model 185 conditionality 4, 43±4, 173±4, 268 agreed 25 alternatives to 289 and CDF 293±5 and con¯ict 59 costs and bene®ts 28 credible 305 the debate 22, 288±90 effectiveness 60±2, 289 ex ante 63 ex post 62±4 externalities 287±8 failure 76, 301±3 hard core and pro forma 289 imposed 25 on incremental spending 199±200 policy 24±6, 27±9, 287±8, 299±312, 300±3 policy-level and policy-change 288±9, 291±2 reasons for 300±1 and SIP 272, 274±5 and time-inconsistency 59±60 trade-offs 290±3 consensus 276, 277±8, 292 see also ownership; participation Consultative Group for International Agricultural Research (CGIAR) 121±2 consumption 219±21 Contingent Credit Line (CCL) 66 cooperative equilibrium 287±8 Coppedge, M. 248 Corbett, J. 66, 68 Cornia, G. A. 182±3 corporate infrastructure 67 corruption 3, 146 cost-ef®ciency 206 CoÃte d'Ivoire 317
country assistance strategies (CASs) 143 country teams 301±2, 305 Craig, S. G. 200±1 crime 93 crisis Asian 68, 70, 72, 102, 268 debt 30, 48, 163±4 ®nancial 65±6, 322±3 management 308 prevention 68±9 relief 30±1, 66, 68 resolution 68±9, 70±1 see also International Monetary Fund Cuba 146 current account de®cits 168±75 Czech Republic 133 de Melo, J. 164, 181, 183, 185 debt crisis 16, 30, 48, 163±4 debt relief 142 debt-overhang effect 316 Deininger, K. 115, 127, 180 Demery, L. 182±3, 185±6 democratic pluralism 4±5 Devarajan, S. 8, 23, 32, 196±208, 267, 288, 291 Development Assistance Committee (DAC) 3, 266 Disch, A. 272 Dollar, D. 8, 27±8, 32, 61±2, 63, 173±4, 176, 178±9, 188, 210±26, 288, 302±4, 310±11, 313±14 donor coordination 206 donor power 304 donor±recipient relations 290, 306±8 Dornbusch, R. 163 Dorosh, P. A. 185 Drahos, P. 52 DreÁze, J. 182 Eade, D. 293 East African Airlines 320 East African Railways 320
Easterly, W. R. 215, 233 Eastern Europe 117 Eaton, J. 285 economic distortions 213±14, 223±4 economic growth aid and 210, 211±16, 224, 311±12 education and 92±4, 100, 182 health services and 92±4, 100, 182 inequality and 89±90, 92 infant mortality and 221 policy conditions for 304 poverty and 118±19, 180 economic growth model 211±16 economic management 118±19, 169, 188 economic rate of return (ERR) 228 and civil liberties 250±1 and policy distortion 234±8, 246 and policy reform 240 re-estimated 232±3 Economic Research Forum 121 economies of scale 118±20, 136 economies of scope 118±20, 285 ECOWATCH 118 education 28, 64±5, 77 and growth 92±4, 100, 182 labour-intensive 65 and policy distortions 240±3 SIP 273 ef®ciency 171±9, 175±9, 187, 188 Eichengreen, B. 72 energy sector reform 176, 178 enforcement 285±7 time-inconsistent 301 see also monitoring environmental factors 117±18, 150 Ethiopia 203, 224, 273 Europe 170, 241 European Union 133, 306, 309 Stabex funds 317 exchange rates and aid 313, 316 ®xed or ¯exible 68±9 Faruqee, R. 186
federal structure 203±4 Feinberg, R. E. 132 Ferreira, F. H. G. 8, 22, 32, 41, 49, 116, 141, 159±93, 199, 283 Feyzioglu, T. 61±2, 200, 220, 224 ®nancial crisis 65±6, 322±3 ®nancial development 66±7 ®nancial management 277±8 ®nancial sector reform 166±7, 176±7 ®nancial stability 276 Financial Times 126 ®scal de®cit 168±75, 234±44 ®scal order 304 Fischer, S. 49, 65, 163, 215 Fishlow, A. 89 ¯ypaper theory 201 foreign exchange reserves 168 Foster, A. 122 Fox, J. A. 133, 148±9 France 133 Freedom House 248, 250±3 fungibility 196±208, 224, 267 consequences of 119, 199±200 de®nition 197±9 research review 200±4 sectoral assistance 201±2 in SIP 272, 274±5 views on 23±4 Furman, J. 1 Gang, I. N. 200 Gardner, R. N. 134 Gavin, M. 15, 19, 54, 66, 282±4 Gaza 138, 147 Germany 22, 67, 133, 145 Gersowitz, M. 285 Ghana 216, 222, 273 Gilbert, C. L. 8, 10±35, 39±82, 282±96 Glewwe, P. 120 global commons problem 44 global public goods 39, 44, 52±3, 57±8, 112±16 Gnanaselvam, S. 143 Gould-Davies, N. 147 government
Index ability to implement reforms 189 accountability 230±1, 313 consumption 220 economic management 118±19, 169, 188 failure 25±6, 39, 43 see also policy; public expenditure; public investment Gramlich, E. 200±1 Greece 133 Green, M. 146 Griesgraber, J. M. 133 Grootaert, C. 95 Grosh, M. 120 Group of Seven (G-7) 148 Group of Three (G-3) 148 Grunberg, I. 52 Guatemala 146 Gunatilaka, R. 267±8 Gunter, B. 133 Gupta, K. L. 200±1 Gwin, C. 20, 133±4, 138 Haggard, S. 25±6 Harberger, A. 72 Harrigan, J. 15, 22, 27, 54, 60±2, 217 Harrold, P. 269±72, 274 Hayter, T. 15 health service and growth 92±4, 100, 182 and policy distortions 240±3 Heavily Indebted Poor Countries (HIPC) 46, 73, 75, 303 Heller, P. S. 183, 200 Henderson, D. 121 Hentschel, J. 93 Herzegovina 139, 147 Heston, A. 162 Hirschman, A. 42, 90 History 17, 70, 95±8 Hoddinott, J. 95 Hopkins, R. 8, 20, 28±9, 32, 54, 132, 282±96, 300 horizontal equity 230 Human Development Report 99±100
Humana 248, 250±3 Hungary 117, 133 Husain, I. 186 Hussein (bib) 2 impact evaluation 114±15 import-substitution policies 161 income distribution evidence on 183±8 inequality 89±90, 92 and structural adjustment 179 income volatility 94±5 India 146±7, 201, 204, 222, 224 Indonesia 66, 70, 99, 146, 167, 189, 222 Programme for Pollution Control, Evaluation and Rating (PROPER) 117±18 industrial sector reform 176, 177 inef®ciency 169 infant mortality 210 and aid 219±23, 224 factors affecting 220±1 and growth 221 in¯ation 168, 181, 216, 304 informational failures 39, 43±4, 51±2 infrastructure corporate and ®nancial 67 roads 269, 273 Inman, R. P. 201 institution-building 167±8 institutional capacity 268 and SIP 272, 275±6 institutions 3 and investment returns 248±53 model 257±8 see also civil liberties insurance 69 interest rate 234±44 International Bank for Reconstruction and Development (IBRD) 12±13, 45±7, 73±4, 287, 292 see also World Bank International Development Agency (IDA) 46, 73, 74±5, 224, 271
International Development Agency (IDA) continued in Bank organisation 12±13, 15, 18 concessional lending 41±2, 47, 48±9, 53, 150 US in¯uence on 138±9 International Finance Corporation (IFC) 12±13, 232, 234, 292, 293 International Financial Institutions (IFIs) 141, 301±2, 304±5 crisis management 308 relations with governments 306±8 International Monetary Fund (IMF) 4, 119, 148, 159, 317, 322 accountability 5 conditionality 24 crisis funding 299±312±300, 301, 306±8 crisis relief 30±1, 40, 70±1 as international lender of last resort 65±6 post-stabilisation role 308±9 relations with World Bank 67, 69, 70±2 style 71±2 US in¯uence 141, 147 investment decisions 229±30 institutional effects 248±53 private 311±12 productivity determinants 235±6 public 229, 231, 244±8 public±private balance 246±7 see also capital Iran 121, 146 Irwin, G. 68 Isham, J. 8, 27±8, 32, 61, 223, 228±63 Italy 133
Jayarajah, C. 169±74, 178, 184±6, 190±2 Jefferson, T. 108 Jha, S. 200, 204 Johnson, J. H. 267, 274 Jolly, R. 183 Jones, S. 23±4, 32, 266±80
Jalan, J. 95 Jamaica 180, 231, 247 James, E. 117 James, H. 147 Japan 133
Laband, D. N. 126 labour markets 167, 181 Latin America 102, 167, 170, 186, 241 Le Prestre, P. 132 lender of last resort, international 65±6
Kahler, M. 26 Kanbur, S. M. R. 7, 17, 31, 41, 52, 87±104, 183, 312±13 Kapur, D. 17, 22±3, 39, 70, 89±90, 96±8, 138, 140, 142, 145±7 Kaufmann, D. 8, 25±8, 32, 61, 223, 228±63 Kaul, I. 52 Kazakhstan 133 Keefer, P. 216 Keely, L. C. 8, 22, 32, 159±93 Kenen, P. 72 Kennan, George 145 Keynes, J. M. 7 Khan, H. A. 200 Khilji, N. M. 200±1 Killick, T. 24, 27, 29, 60, 267±8, 274, 288±90, 292 Kindleberger, C. P. 14 Kinsey, W. 95 Knack, S. 216 knowledge 49±51, 57 public good 108, 112±16 undersupply 108 see also advice Kohl, H. 98 Korea 66, 70, 167 Krueger, A. O. 25, 141, 175, 283 Krugman, P. 42, 66 Kuznets, S. 89 curve 91
Index lending 145±7 headroom 31 instruments 205±7 MIGA 47, 48±9 multilateral 287 non-repayment 301±2 policy and project based 21±3, 27 pooled risk 136 programmes 187 SALS 21±2, 24, 30, 159, 169, 170±5, 183 SECALS 21±2, 24, 160, 183 split from monitoring 67 SSAL 75 see also aid Lessard, D. R. 283 leverage 206 Lewis, L. P. 17, 22±3, 39, 70, 89±90, 96±8, 138, 140, 142, 145±7 Li, H. 92 liquidity 68±9 Lister, F. K. 133±4 Litch®eld, J. A. 181, 187 Little, I. M. D. 90, 161 Living Standards Measurement Study (LSMS) 120 Lockwood, M. 290 Loxley, J. 121 Lucas, R. 182 Lustig, N. 92, 94±5, 186 Luxembourg 133 McCloy, John 134 McGuire, M. C. 200±1 McNamara, Robert 15, 18, 87, 96, 98, 135, 140 Macroeconomic Adjustment Programmes (MAPs) 71 macroeconomic shocks 91 Malta 133 Mankiw, N. G. 233 market failure 2, 42±4 market liberalisation 100±1, 102 Marr, A. 267±8 Marshall Plan 145
Mason, E. S. 13±14 Meade, J. 92 Mexican crisis 30 Mexico 47, 69, 117, 118, 139, 141, 146, 147, 163 Meyer, E. 134 Middle East 146, 170 Mieszkowski, P. 200 Mikesell, R. F. 133 Miller, M. 137 monitoring 56 split from lending 67 see also enforcement Morawetz, D. 15 Morduch, J. 95 Morley, S. 186 Morris, S. 305 Morrison, K. 100 Morrisson, C. 164, 183, 185 Morrow, J. D. 135 Mosley, P. 15, 22, 27, 54, 60±2, 217 Mozambique 273 Multinational Investment Guarantee Agency (MIGA) 12±13, 47±9, 293 Murphy, R. 42 Naim, M. 39, 143, 282 National Science Foundation 124 Nelson, J. 143 Nelson, P. 133, 149 Nepal, Arun III Dam 148 NGOs 25, 29, 41, 54, 77, 148±51, 290, 293 Nicaragua 145 Nigria 321 NoguaÄs, J. 49 Nordic policy 314 North Africa 170, 241 Oakland, W. 200 OAU 322 O'Connell, S. A. 234, 305 OECD 119 of®cial development assistance (ODA) 266
of®cial development assistance (ODA) continued see also aid Ogata, S. 133±4 oil price shocks 161±3 Oliver, R. W. 14, 39, 282 openness 94, 114, 216, 304 Operations Evaluation Department (OED) 61, 171±4, 177±8, 185, 232, 241 ownership 3, 102, 143, 178±9, 267, 277 and conditionality 59, 62±3, 290 lack of 27±8 and SIP 272, 273±4 see also consensus; participation Oye, K. 135 Pack, H. and J. R. 200±2 Pakistan 317 Palestine 139 participation 3, 143±4, 229, 230 see also consensus; ownership partnership 307, 322±3 Pattillo, C. 302 pension reform 116±17 Philippines 118, 146, 231 Picciotto, R. 143 Piette, M. J. 126 Polak, J. 70 Poland 117 policy, see also government policy conditionality see conditionality policy distortions combined effect 235, 238±9 and education 240±3 and ERR 234±8, 246 and health 240±3 model 257±8 policy reform 62±3 analysis of 114 and ERR 240 policy regime 228±9 and aid effectiveness 267 good 28 and growth 304
and SIP 271±2, 273 political stability 174 poor people characteristics of 99 short- and long-term effects on 188 subgroups of 187 Portugal 133 poverty and growth 118±19, 180 indicators 186±7 poverty reduction 53 changing theory of 88±95 changing World Bank views 16±19, 95±101 through capacity building 41±2 through growth 88±9 trickle-down view 19, 121, 133, 146 Powell, A. 7±8, 31, 39±82, 132, 282±96 Prebisch, R. 89 prices 175±6 distortion 230, 234±44 reform 166 relative 180±1 principal±agent problem 50±1, 109, 116 Pritchett, L. H. 221, 223, 251±2, 254 private banking 287±8 private sector borrowing 47±8 privatisation 22, 167 project performance choice of country characteristics 233±4 and civil strife 251±3 evaluation 114 failure 223±4 inef®cient 128 policy determinants 232±44 see also aid ; economic rate of return; investment returns; lending projects 312 choice of 253±7 selectivity 30, 255±6, 288±9, 290, 294 Psacharopoulos, G. 186 public enterprise reform 176, 177
Index public expenditure 276 effects on poverty 181±2 public expenditure reform loan (PERL) 200, 206±7, 276 Public Expenditure Review 268 public goods advice as 76 global 39, 44, 52±3, 57±8, 112±16 knowledge 108, 112±16 public investment 229, 231, 244±8 Purcell, J. 137 Radelet, S. 66 Rajkumar, A. S. 202±3, 202±3 Ranis, G. 139 Rapkin, D. 133 Ravallion, M. 92, 95, 118, 180 Rawls, J. 41 Reagan, R. 97±8, 141 Rebelo, S. T. 215 redistribution, global 41±2 reforms microeconomic 167±8 performance 171±5 timing and sequencing 169, 179, 189 regional coordination 320±2 Regional Research Consortia 121 Reinicke, W. H. 52, 248 research 108±29, 139±44 arguments for 112±22 budget 49, 111±12, 113 impact 124±7 in-house champion 109, 116±18 level of 110 output 124±5 outsourced 120±2 production technology 118±20 quality of 123±4 risk income volatility 94±5 loans 136 policy 56 Road Maintenance Initiative (RMI) 269, 273
Rodrik, D. 1, 15, 19, 54, 94, 142, 282±4, 287 Romer, D. 233 Rosen, H. 200 Rosenstein-Rodan, P. 23, 42, 88 Rosenzweig, M. 122 Roy, A. 8, 282±96 Russia 147, 159, 167 Sachs, J. D. 66, 94, 161±3, 215 Sahn, D. E. 185 Sala-i-Martin, X. 234 Sandler, T. 52, 100 Schmidt-Hebbel, K. 111 Scitovsky, T. 90 Scott, M. 90 Sector Adjustment Strategies (SASs) 23 Sector Investment Programme (SIP) 266, 268±73, 305 in Africa 273±6 and aid effectiveness 276±8 Bank role 271 de®ned 270±1 preconditions for 271 Sector-Wide Approaches (SWAps) 266 Sectoral Adjustment Loans (SECALs) 21±2, 24, 160, 183 replaced SALs 171±2 reviewed 176±9 sectoral concessionary loans 202 Sen, A. 182 Sen, B. 170±4, 178, 185±6, 190±1 Serven, L. 111 Shihata, I. 148 Shleifer, A. 42 Snidal, D. 135 social services 187 Solow, R 233 Sommer, D. 134 Soviet Union 138, 139, 145 Special Program of Assistance to Africa 300 Special Stuctural Adjustment Loan (SSAL) 75
Squire, L. 7, 19±20, 31, 49±51, 92, 108±29, 139, 180, 185±6, 200, 204 stabilisation 166 concept 160±1 policies 168±75 Stein, A. 135 Stern, E. 98 Stern, M. 52 Stern, N. 41, 49, 116, 141, 283 Stevens, M. 143 Stewart, F. 182±3 Stiglitz, J. E. 1±8, 17, 29±30, 49, 70, 72, 102, 142, 285 Strand, J. 133 Strategic Compact 144 structural adjustment 1980s 16, 91, 97±8, 159±93 causes of 161±4 concept 160±1 and equity 179±88 failure 180, 183±5, 188 increased lending for 170±1 nature of 164±8 policy reforms 166±7 relative prices effect 180±1 Structural Adjustment Loans (SALS) 21±2, 24, 30, 159, 169, 183 reviewed 170±5 Sub-Saharan Africa 202±3, 303, 320 Sudan 231, 247 Summers, L. 31, 71, 221 Summers, R. 162 supply side 175±6 Suthiwart-Narueput, S. 200, 204 Svensson, J. 61, 173±4, 176, 178, 188, 302 Swaroop, V. 8, 23, 32, 61±2, 196±208, 220, 224, 267, 288, 291 Tanzania 186, 216 Integrated Road Project (TIRP) 269 taxation and government accountability 313 reform 166
revenues 317±19 Taylor, L. 143 Thailand 66, 70, 118, 167, 216, 224 Thorbecke, E. 121 Tower, E. 127 Toye, J. 15, 22, 27, 54, 60±2, 217 trade liberalisation 22 policies 181 reform 166 restriction 234±44 transitional dynamics 169 transparency 230±1 trickle-down 87, 91±2, 97 Turkey 19, 121, 133, 146 Uganda 180, 273, 309, 317, 320 uncertainty 6 United Kingdom 22, 39, 67, 133, 140, 282, 300, 303 United Nations (UN) 119, 224, 303, 306, 322 Special Programme for Africa (SPA) 266 UNDP 5 United States 39, 49, 140, 282 Congress 151 education expenditure 201 Federal Reserve 141 in¯uence on World Bank 20±2, 133±7 leverage on IDA 138±9 oil price shocks 163 Treasury 141, 146±7 USAID 314 university courses 126±7 Uruguay 117 Uzawa, H. 182 Vietnam 93, 138, 224 Vines, D. 7, 10±35, 39±82, 87±104 virtuous circle 88 Vishny, R. 42 Wade, R. 10, 149±50
Index Walters, A. 10, 284 Wang, Y. 243±4 Wapenhans Report 269, 275 Warner, A. 94, 215 Washington consensus 1, 16±17, 19, 49, 87, 91±2, 100, 141±3 Wasty, S. S. 267, 274 Weaving, R. 143 Webb, R. 17, 22±3, 39, 70, 89±90, 96±8, 138, 140, 142, 145±7 Weil, D. N. 233 welfare dependency 312±13, 314±16 White, H.D. 14±15 Williamson, J. 16±17, 100 Wise, L. R. 124 Wolfensohn, James D. 6, 17, 29, 40±1, 52, 76±7, 101±2, 128, 148, 207, 293 Wood, A. 290 Woods, N. 7, 21, 32, 132±52 World Bank Articles of Agreement 74, 136, 145, 150±1 banking role 283, 285±7 as Conditionality Bank 54, 55±6, 58, 59±65 data base 116, 119±20 development agency 14, 51±3, 70±1, 283±7 Economic and Sector Work (EDW) 49, 111, 127 ®nances 45±6, 74, 137±9, 150 formation 145 function additionality 51, 53±5, 285± 8, 295 functional analysis 282±4 future for 11 governance 5±6, 20±1, 133±4 growth of 15 history 14±17 inspection panel 148
institution 56, 64±5, 72±8 institutional organisation 12±13 Knowledge Bank 10±11, 28±30, 40, 55, 57±8, 64±5, 72±3, 77, 128, 144, 311 lending institution 31, 45±9, 145±7 level of autonomy 137±47 multilateral 135±7 objectives 10 Operations Evaluation Department (OED) 61, 171±4, 177±8, 185, 232, 242 policy spectrum 292±3 pricing policy review75 rationale for 39, 41±4 relations with IMF 67, 69, 70±2 reorganisation 76 Research Advisory Staff 125 research institute 19±20, 40, 49±51, 108±29, 283±4 Research Support 124 reserves 31, 74±6 structures 73±4 style 71±2 technical expertise 139±44 use of English 140±1 World Development Report (WDR) 99±101, 125±6, 160, 183 World Trade Organisation (WTO) 119, 136 Wriston, W. 162 Yugoslavia 145 Zaire 217, 222 Zaman, H. 93 Zambia 216, 222, 273, 274±5 Zampelli, E. M. 200±1 Zhu, M. 61±2, 199±200, 202, 220, 224 Zou, H. F. 92